Is it music?
Pauline Larrouy-Maestri (Exp 1-3) & T. Ata Aydin (Exp 4)
Experiment 1
Analysis of the Music data collected in September 2022
Following the preregistration: Larrouy-Maestri, P., & Wald-Fuhrmann, M. (2022, September 12). Definition of music: Effect of self- vs others-reference in the identification of auditory stimuli as music or not. Retrieved from osf.io/sb24q
Clear and load
install.packages("vcd")
trying URL 'https://cran.rstudio.com/bin/macosx/big-sur-x86_64/contrib/4.5/vcd_1.4-13.tgz'
Content type 'application/x-gzip' length 1304651 bytes (1.2 MB)
==================================================
downloaded 1.2 MB
The downloaded binary packages are in
/var/folders/ry/qb5hyyns70q661g47x8kv7hx3txcp3/T//RtmpTf2wYm/downloaded_packagesRescale the Confidence ratings(0-3 instead of 1-4) to match the intention and the methods
# Create a new column 'confidence_rescaled' that is a copy of 'Confidence'
Alldata$confidence_rescaled <- Alldata$Confidence
# Rescale 'confidence_rescaled' for all data
Alldata$confidence_rescaled<- Alldata$confidence_rescaled - 1ANALYSIS 1: Effect of Condition (self versus other) on the categorisation of stimuli
Mixed effects logistic regression will be applied to predict the response (i.e., “Music” or “Not music”) from the condition, with random intercepts for participants and stimuli items.
NOTE that the syntax in the registration included the “a priori” condition but is not presented in the paper (only mentioned as a footnote) Proposed tentative syntax (in the registration): Music_answer ~ a priori group + Condition + a priori group:Condition + (1|Participant) + (1|Stim)
Code to change the baseline of the factors (self) for the model
Alldata$Condition <- as.factor(Alldata$Condition)
Alldata$Condition <- relevel(Alldata$Condition, ref = "self")Null and simplest model, including only Condition:
Effect_null <- glmer(data = Alldata, Music_answer ~ 1 + (1|Participant) + (1|Stim), family = binomial)
summary(Effect_null)Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
Family: binomial ( logit )
Formula: Music_answer ~ 1 + (1 | Participant) + (1 | Stim)
Data: Alldata
AIC BIC logLik -2*log(L) df.resid
8906.7 8930.1 -4450.4 8900.7 18081
Scaled residuals:
Min 1Q Median 3Q Max
-40.002 -0.179 0.030 0.159 16.533
Random effects:
Groups Name Variance Std.Dev.
Participant (Intercept) 0.9493 0.9743
Stim (Intercept) 18.6233 4.3155
Number of obs: 18084, groups: Participant, 202; Stim, 90
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 0.8645 0.4658 1.856 0.0635 .
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1Effect_log <- glmer(data = Alldata, Music_answer ~ Condition + (1|Participant) + (1|Stim), family = binomial)
summary(Effect_log)Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
Family: binomial ( logit )
Formula: Music_answer ~ Condition + (1 | Participant) + (1 | Stim)
Data: Alldata
AIC BIC logLik -2*log(L) df.resid
8903.1 8934.3 -4447.6 8895.1 18080
Scaled residuals:
Min 1Q Median 3Q Max
-39.331 -0.177 0.031 0.161 16.680
Random effects:
Groups Name Variance Std.Dev.
Participant (Intercept) 0.9208 0.9596
Stim (Intercept) 18.6715 4.3211
Number of obs: 18084, groups: Participant, 202; Stim, 90
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 1.0413 0.4715 2.209 0.0272 *
Conditionother -0.3488 0.1464 -2.383 0.0172 *
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Correlation of Fixed Effects:
(Intr)
Conditinthr -0.154anova(Effect_null,Effect_log)Data: Alldata
Models:
Effect_null: Music_answer ~ 1 + (1 | Participant) + (1 | Stim)
Effect_log: Music_answer ~ Condition + (1 | Participant) + (1 | Stim)
npar AIC BIC logLik -2*log(L) Chisq Df Pr(>Chisq)
Effect_null 3 8906.7 8930.1 -4450.4 8900.7
Effect_log 4 8903.1 8934.3 -4447.6 8895.1 5.5878 1 0.01809 *
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1Details about the full model
tidy (Effect_log)performance(Effect_log)# Indices of model performance
AIC | AICc | BIC | R2 (cond.) | R2 (marg.) | ICC | RMSE | Sigma | Log_loss | Score_log | Score_spherical
-------------------------------------------------------------------------------------------------------------------------
8903.136 | 8903.138 | 8934.347 | 0.856 | 0.001 | 0.856 | 0.257 | 1.000 | 0.217 | -Inf | 2.285e-04ICC for the random effects (all together and separately)
# ICC calculation by group
icc_results_group <- performance::icc(Effect_log, by_group = TRUE)
print(icc_results_group)# ICC by Group
Group | ICC
-------------------
Participant | 0.040
Stim | 0.816# ICC calculation for all random
icc_results_all <- performance::icc(Effect_log, by_group = FALSE)
print(icc_results_all)# Intraclass Correlation Coefficient
Adjusted ICC: 0.856
Unadjusted ICC: 0.855ANALYSIS 1bis: Effect of condition (Condition: self versus other) on the confidence ratings (confidence as ordinal variable)
conf_ord_r <- mutate(Alldata, confidence_rescaled = factor(confidence_rescaled, levels=0:3, ordered=TRUE))
conf_ord_r <- mutate(Alldata, Condition = factor(Condition, ordered=FALSE))
# Check the class and levels of the 'Confidence' variable
class(conf_ord_r$confidence_rescaled)[1] "numeric"levels(conf_ord_r$confidence_rescaled)NULL# Convert 'Confidence' to factor if it's not already
conf_ord_r$confidence_rescaled <- factor(conf_ord_r$confidence_rescaled, ordered = TRUE)
# Check the class and levels again
class(conf_ord_r$confidence_rescaled)[1] "ordered" "factor" levels(conf_ord_r$confidence_rescaled)[1] "0" "1" "2" "3"# Fit the CLMM model without the effect of Condition
Effect_Conf_null_r <- clmm(confidence_rescaled ~ 1 + (1|Participant) + (1|Stim), data = conf_ord_r, link = "probit", threshold = "equidistant")
summary(Effect_Conf_null_r)Cumulative Link Mixed Model fitted with the Laplace approximation
formula: confidence_rescaled ~ 1 + (1 | Participant) + (1 | Stim)
data: conf_ord_r
Random effects:
Groups Name Variance Std.Dev.
Participant (Intercept) 0.2899 0.5384
Stim (Intercept) 0.7301 0.8545
Number of groups: Participant 202, Stim 90
No Coefficients
Threshold coefficients:
Estimate Std. Error z value
threshold.1 -2.59734 0.10010 -25.95
spacing 1.13614 0.01139 99.77# Fit the CLMM model with the effect of Condition
model_conf_ord_r <- clmm(confidence_rescaled ~ Condition + (1 | Participant) + (1 | Stim), data = conf_ord_r, link = "probit", threshold = "equidistant")
summary(model_conf_ord_r)Cumulative Link Mixed Model fitted with the Laplace approximation
formula: confidence_rescaled ~ Condition + (1 | Participant) + (1 | Stim)
data: conf_ord_r
Random effects:
Groups Name Variance Std.Dev.
Participant (Intercept) 0.2894 0.5379
Stim (Intercept) 0.7301 0.8544
Number of groups: Participant 202, Stim 90
Coefficients:
Estimate Std. Error z value Pr(>|z|)
Conditionother -0.04595 0.07816 -0.588 0.557
Threshold coefficients:
Estimate Std. Error z value
threshold.1 -2.62030 0.10746 -24.38
spacing 1.13614 0.01139 99.77# Comparison of the null and full model for confidence
anova(Effect_Conf_null_r,model_conf_ord_r)Likelihood ratio tests of cumulative link models:
no.par AIC logLik LR.stat df Pr(>Chisq)
Effect_Conf_null_r 4 29849 -14921
model_conf_ord_r 5 29851 -14920 0.3454 1 0.5568# ICC calculation by group (i.e., each random effects) for the NULL model (since the full model is not better)
icc_results_group <- performance::icc(Effect_Conf_null_r, by_group = TRUE)
print(icc_results_group)# ICC by Group
Group | ICC
-------------------
Participant | 0.144
Stim | 0.361# ICC calculation for all random effects together for the NULL model (since the full model is not better)
icc_results_all <- performance::icc(Effect_Conf_null_r, by_group = FALSE)
print(icc_results_all)# Intraclass Correlation Coefficient
Adjusted ICC: 0.505
Unadjusted ICC: 0.505Figure 1A
# Compute mean music responses and mean confidence ratings
DataPlot <- with(Alldata, aggregate(cbind(Music_answer, confidence_rescaled), list(Stimuli=Stim, Condition = Condition), mean))
DataPlotSetting colors for the plots
self = "orange"
other = "purple"
Condition_colors = c(self,other)Plotting the music answers with 2 contrasted colors (Conditions: 1st and 3rd person, also names Self and Other conditions), with order according to the Self values
# Calculate the mean Music_answer for each stimulus in the "self" condition
self_means <- with(subset(Alldata, Condition == "self"),
tapply(Music_answer, Stim, mean, na.rm = TRUE))
# Ensure the Music_answer column is numeric
DataPlot$Music_answer <- as.numeric(DataPlot$Music_answer)
# Reorder the Stimuli factor levels based on the means in the "self" condition
DataPlot$Stimuli <- factor(DataPlot$Stimuli,
levels = names(sort(self_means, decreasing = FALSE)))
# Reorder the levels of the Condition factor
DataPlot$Condition <- factor(DataPlot$Condition, levels = c("self", "other"))
ResponsePlot <- ggplot(DataPlot, aes(x = reorder(Stimuli, Music_answer), y = Music_answer, group = Condition, colour = Condition)) +
geom_point(aes(shape = Condition), size = 3.5) +
scale_colour_manual(values = Condition_colors,
labels = c("1st person", "3rd person")) + # Rename Condition labels
scale_shape_manual(values = c("self" = 16, # Solid circle
"other" = 1), # Hollow circle
labels = c("1st person", "3rd person")) +
theme_classic() +
xlab("Stimuli (n = 90)") +
ylab("Music responses") +
theme(axis.text.x = element_blank(),
legend.text = element_text(size = 10),
legend.title = element_text(size = 10),
legend.position = c(0.8, 0.4),
axis.title.y = element_text(size = 14),
axis.title.x = element_text(size = 12),
panel.background = element_rect(fill = "white")) # Ensure white background
ResponsePlot
Plotting Confidence for self and other conditions
# Reorder the Stimuli factor levels based on the means in the "self" condition
DataPlot$Stimuli <- factor(DataPlot$Stimuli,
levels = names(sort(self_means, decreasing = FALSE)))
# Reorder the levels of the Condition factor
DataPlot$Condition <- factor(DataPlot$Condition, levels = c("self", "other"))
ResponsePlot_conf <- ggplot(DataPlot, aes(x = reorder(Stimuli, Music_answer), y = confidence_rescaled, group = Condition, colour = Condition)) +
geom_point(aes(shape = Condition), size = 3.5) + # Adjust size for circles
scale_colour_manual(values = Condition_colors,
labels = c("1st person", "3rd person")) + # Rename Condition labels
scale_shape_manual(values = c("self" = 16, # Solid circle for "self"
"other" = 1), # Hollow circle for "other"
labels = c("1st person", "3rd person")) +
theme_classic() +
xlab("Stimuli (n = 90)") +
ylab("Confidence") +
ylim(1.3, 3) + # Adjust y-axis limits as needed
theme(axis.text.x = element_blank(),
legend.text = element_text(size = 12), # Change legend text font size
legend.title = element_text(size = 12), # Change legend title font size
legend.position = "none", # Remove the legend
axis.title.y = element_text(size = 14),
axis.title.x = element_text(size = 12)) +
theme(panel.background = element_rect(fill = "white"))
ResponsePlot_conf
NA
NACombine plots Identification/Confidence and save
Figure_1 <- grid.arrange(ResponsePlot,ResponsePlot_conf,ncol = 1)
ggsave("~/Desktop/ResponsePlot_exp1.tiff", plot = Figure_1, width = 5, height = 5, dpi = 600, device = "tiff", compression = "lzw")
Experiment 2
Analysis of the Music data collected in November 2022
Following the preregistration: Larrouy-Maestri, P., & Wald-Fuhrmann, M. (2022, November 21). Perception of sounds as music or not: effect of stimulus duration Retrieved from osf.io/azkcp
Clear and load
rm(list = ls())
# Set working directory
setwd("/Users/p.larrouy/Desktop/Music_osf")Error in setwd("/Users/p.larrouy/Desktop/Music_osf") :
cannot change working directoryRescale the Confidence ratings for duration short and long (0-3 instead of 1-4) to match the intention, the methods, and the date of the medium duration (i.e., data of Exp1, self)
# Create a new column 'confidence_rescaled' that is a copy of 'confidence'
Alldata$confidence_rescaled <- Alldata$confidence
# Rescale 'confidence_rescaled' for Duration values 2 and 9
Alldata$confidence_rescaled[Alldata$Duration_label %in% c("short", "long")] <- Alldata$confidence_rescaled[Alldata$Duration_label %in% c("short", "long")] - 1
# Optional: Verify the changes
# You can check a subset of the data to ensure the changes are correct
head(Alldata[Alldata$Duration_label %in% c("short", "long"), c("Duration_label", "confidence", "confidence_rescaled")])NAANALYSIS 1: Effect of duration (2 versus 5 versus 10) on the categorisation of stimuli
Mixed effects logistic regression will be applied to predict the response (i.e., “Music” or “Not music”) from the condition (i.e., duration: 2, 5, 10 seconds), with random intercepts for participants and stimuli items.
NOTE that we don’t use the results of Exp1 (3 groups:music, no_music, ambiguous) since we reuse the data of the “self” group from which the three groups were defined
NOTE that the syntax in the registration included the “a priori” condition but is not presented in the paper (only mentioned as a footnote) Proposed tentative syntax (in the registration): Music_answer ~ a priori group + Duration + a priori group:Duration + (1|exp_subject_id) + (1|stim_name)
Code to change the baseline of the factors (Duration_label) for the model
Alldata$Duration_label <- as.factor(Alldata$Duration_label)
Alldata$Duration_label <- relevel(Alldata$Duration_label, ref = "medium")Null and simplest model, including only duration:
Effect_null <- glmer(data = Alldata, Music_answer ~ 1 + (1|exp_subject_id) + (1|stim_name), family = binomial)
summary(Effect_null)Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
Family: binomial ( logit )
Formula: Music_answer ~ 1 + (1 | exp_subject_id) + (1 | stim_name)
Data: Alldata
AIC BIC logLik -2*log(L) df.resid
11002 11026 -5498 10996 22347
Scaled residuals:
Min 1Q Median 3Q Max
-18.4936 -0.1994 0.0281 0.1433 19.6758
Random effects:
Groups Name Variance Std.Dev.
exp_subject_id (Intercept) 0.9988 0.9994
stim_name (Intercept) 20.2061 4.4951
Number of obs: 22350, groups: exp_subject_id, 298; stim_name, 75
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 1.6566 0.5289 3.132 0.00174 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1Effect_log <- glmer(data = Alldata, Music_answer ~ Duration_label + (1|exp_subject_id) + (1|stim_name), family = binomial)
summary(Effect_log)Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
Family: binomial ( logit )
Formula: Music_answer ~ Duration_label + (1 | exp_subject_id) + (1 | stim_name)
Data: Alldata
AIC BIC logLik -2*log(L) df.resid
10993.3 11033.3 -5491.6 10983.3 22345
Scaled residuals:
Min 1Q Median 3Q Max
-18.1067 -0.1984 0.0287 0.1436 20.0522
Random effects:
Groups Name Variance Std.Dev.
exp_subject_id (Intercept) 0.9486 0.9739
stim_name (Intercept) 20.1970 4.4941
Number of obs: 22350, groups: exp_subject_id, 298; stim_name, 75
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 1.9430 0.5345 3.635 0.000278 ***
Duration_labellong -0.3282 0.1509 -2.175 0.029660 *
Duration_labelshort -0.5396 0.1508 -3.579 0.000345 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Correlation of Fixed Effects:
(Intr) Drtn_lbll
Drtn_lbllng -0.139
Drtn_lblshr -0.139 0.493 anova(Effect_null,Effect_log)Data: Alldata
Models:
Effect_null: Music_answer ~ 1 + (1 | exp_subject_id) + (1 | stim_name)
Effect_log: Music_answer ~ Duration_label + (1 | exp_subject_id) + (1 | stim_name)
npar AIC BIC logLik -2*log(L) Chisq Df Pr(>Chisq)
Effect_null 3 11002 11026 -5498.0 10996
Effect_log 5 10993 11033 -5491.6 10983 12.697 2 0.00175 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1Details about the full model
tidy (Effect_log)performance(Effect_log)# Indices of model performance
AIC | AICc | BIC | R2 (cond.) | R2 (marg.) | ICC | RMSE | Sigma | Log_loss | Score_log | Score_spherical
----------------------------------------------------------------------------------------------------------------------------
10993.270 | 10993.273 | 11033.343 | 0.866 | 0.002 | 0.865 | 0.259 | 1.000 | 0.219 | -Inf | 6.644e-05ICC for the random effects (all together and separately)
# ICC calculation by group
icc_results_group <- performance::icc(Effect_log, by_group = TRUE)
print(icc_results_group)# ICC by Group
Group | ICC
----------------------
exp_subject_id | 0.039
stim_name | 0.827# ICC calculation for all random
icc_results_all <- performance::icc(Effect_log, by_group = FALSE)
print(icc_results_all)# Intraclass Correlation Coefficient
Adjusted ICC: 0.865
Unadjusted ICC: 0.864ANALYSIS 1bis: Effect of condition (duration: short, medium, long) on the confidence ratings (confidence as ordinal variable)
conf_ord_r <- mutate(Alldata, confidence_rescaled = factor(confidence_rescaled, levels=0:3, ordered=TRUE))
conf_ord_r <- mutate(Alldata, Condition = factor(Duration_label, ordered=FALSE))
# Check the class and levels of the 'Confidence' variable
class(conf_ord_r$confidence_rescaled)[1] "numeric"levels(conf_ord_r$confidence_rescaled)NULL# Convert 'Confidence' to factor if it's not already
conf_ord_r$confidence_rescaled <- factor(conf_ord_r$confidence_rescaled, ordered = TRUE)
# Check the class and levels again
class(conf_ord_r$confidence_rescaled)[1] "ordered" "factor" levels(conf_ord_r$confidence_rescaled)[1] "0" "1" "2" "3"# Fit the CLMM model without the effect of Condition
Effect_Conf_null_r <- clmm(confidence_rescaled ~ 1 + (1|exp_subject_id) + (1|stim_name), data = conf_ord_r, link = "probit", threshold = "equidistant")
summary(Effect_Conf_null_r)Cumulative Link Mixed Model fitted with the Laplace approximation
formula: confidence_rescaled ~ 1 + (1 | exp_subject_id) + (1 | stim_name)
data: conf_ord_r
Random effects:
Groups Name Variance Std.Dev.
exp_subject_id (Intercept) 0.3114 0.558
stim_name (Intercept) 0.7465 0.864
Number of groups: exp_subject_id 298, stim_name 75
No Coefficients
Threshold coefficients:
Estimate Std. Error z value
threshold.1 -2.56205 0.10664 -24.02
spacing 1.11763 0.01006 111.13# Fit the CLMM model with the effect of Condition
model_conf_ord_r <- clmm(confidence_rescaled ~ Duration_label + (1 | exp_subject_id) + (1 | stim_name), data = conf_ord_r, link = "probit", threshold = "equidistant")
summary(model_conf_ord_r)Cumulative Link Mixed Model fitted with the Laplace approximation
formula: confidence_rescaled ~ Duration_label + (1 | exp_subject_id) + (1 | stim_name)
data: conf_ord_r
Random effects:
Groups Name Variance Std.Dev.
exp_subject_id (Intercept) 0.3104 0.5572
stim_name (Intercept) 0.7465 0.8640
Number of groups: exp_subject_id 298, stim_name 75
Coefficients:
Estimate Std. Error z value Pr(>|z|)
Duration_labellong 0.01056 0.08187 0.129 0.897
Duration_labelshort -0.05704 0.08164 -0.699 0.485
Threshold coefficients:
Estimate Std. Error z value
threshold.1 -2.57749 0.11639 -22.14
spacing 1.11763 0.01006 111.13# Comparison of the null and full model for confidence
anova(Effect_Conf_null_r,model_conf_ord_r)Likelihood ratio tests of cumulative link models:
no.par AIC logLik LR.stat df Pr(>Chisq)
Effect_Conf_null_r 4 37154 -18573
model_conf_ord_r 6 37157 -18573 0.7844 2 0.6756# ICC calculation by group (i.e., each random effects) for the NULL model (since the full model is not better)
icc_results_group <- performance::icc(Effect_Conf_null_r, by_group = TRUE)
print(icc_results_group)# ICC by Group
Group | ICC
----------------------
exp_subject_id | 0.151
stim_name | 0.363# ICC calculation for all random effects together for the NULL model (since the full model is not better)
icc_results_all <- performance::icc(Effect_Conf_null_r, by_group = FALSE)
print(icc_results_all)# Intraclass Correlation Coefficient
Adjusted ICC: 0.514
Unadjusted ICC: 0.514Figure 1B
# Compute mean music responses and mean confidence ratings
DataPlot <- with(Alldata, aggregate(Music_answer, list(Stimuli=stim_name,Condition=Duration_label), mean))Setting colors for the plots
s = "gold"
m = "orange"
l = "brown"
dur_colors = c(s,m,l)Plotting the music answers with 3 colors (Conditions: Short, Medium, and Long stimuli)
DataPlot <- with(Alldata, aggregate(Music_answer, list(Stimuli=stim_name,Condition=Duration_label), mean))
# Reorder the levels of the Condition factor
DataPlot$Condition <- factor(DataPlot$Condition, levels = c("short", "medium", "long"))
# Create the plot with the reordered Condition levels
ResponsePlot <- ggplot(DataPlot, aes(x = reorder(Stimuli, x), y = x, group = Condition, colour = Condition)) +
geom_point(aes(shape = Condition), size = 3.5) +
scale_colour_manual(values = dur_colors,
labels = c("Short", "Medium", "Long")) +
scale_shape_manual(values = c("short" = 1, # Solid circle
"medium" = 16,
"long"= 1), # Hollow circle
labels = c("Short", "Medium", "Long")) +
theme_classic() +
xlab("Stimuli (n = 75)") +
ylab("Music responses") +
theme(axis.text.x = element_blank(),
legend.text = element_text(size = 10),
legend.title = element_text(size = 10),
legend.position = c(0.8, 0.4),
axis.title.y = element_text(size = 14),
axis.title.x = element_text(size = 12)) +
theme(panel.background = element_rect(fill = "white"))
ResponsePlot
Plotting Confidence for Short, Medium, and Long stimuli
# Compute mean music responses and mean confidence ratings
DataPlot <- with(Alldata, aggregate(cbind(Music_answer, confidence_rescaled), list(Stimuli=stim_name, Condition = Duration_label), mean))
DataPlot
# Reorder the levels of the Condition factor
DataPlot$Condition <- factor(DataPlot$Condition, levels = c("short", "medium", "long"))
# Plot confidence like for identification
ResponsePlot_conf <- ggplot(DataPlot, aes(x = reorder(Stimuli, Music_answer), y = confidence_rescaled, group = Condition, colour = Condition)) +
geom_point(aes(shape = Condition), size = 3.5) +
scale_colour_manual(values = dur_colors,
labels = c("Short", "Medium", "Long")) +
scale_shape_manual(values = c("short" = 1, # Hollow circle
"medium" = 16, # Solid circle
"long"= 1), # Hollow circle
labels = c("Short", "Medium", "Long")) +
theme_classic() +
# scale_shape(solid = FALSE) +
xlab("Stimuli (n = 75)") +
ylab("Confidence") +
ylim(1.3, 3) + # Adjust y-axis limits as needed
theme(axis.text.x = element_blank(),
legend.text = element_text(size = 12), # Change legend text font size
legend.title = element_text(size = 12), # Change legend title font size
legend.position = "none", # Remove the legend
axis.title.y = element_text(size = 14),
axis.title.x = element_text(size = 12)) +
theme(panel.background = element_rect(fill = "white"))
ResponsePlot_conf
Combine plots and save
Figure_1B <- grid.arrange(ResponsePlot,ResponsePlot_conf,ncol = 1)
ggsave("~/Desktop/ResponsePlot_exp2.tiff", plot = Figure_1B, width = 4, height = 5, dpi = 600, device = "tiff", compression = "lzw")
Experiment 3
Analysis of the Music data collected in November 2022
Following the preregistration: Larrouy-Maestri, P., & Wald-Fuhrmann, M. (2022, November 21). Perception of sounds as music or not: effect of stimulus duration Retrieved from osf.io/azkcp
Clear and load
rm(list = ls())
# Set working directory
setwd("/Users/p.larrouy/Desktop/Music_osf")Error in setwd("/Users/p.larrouy/Desktop/Music_osf") :
cannot change working directoryANALYSIS 1: Effect of repetition on the categorisation of stimuli
Mixed effects logistic regression will be applied to predict the response (i.e., “Music” or “Not music”) from the condition (i.e., first or second presentation), with random intercepts for participants and stimuli items.
NOTE that the syntax in the registration included the “a priori” condition but is not presented in the paper (only mentioned as a footnote) Proposed tentative syntax (in the registration): Music_answer ~ a priori group + Order + a priori group:Order + (1|exp_subject_id) + (1|stim_name)
Code to change the baseline of the factors (Order_letter) for the model
Alldata$Order_letter <- as.factor(Alldata$Order_letter)
Alldata$Order_letter <- relevel(Alldata$Order_letter, ref = "test")Null and simplest model, including only duration:
Effect_null <- glmer(data = Alldata, Music_answer ~ 1 + (1|exp_subject_id) + (1|stim_name), family = binomial)
summary(Effect_null)Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
Family: binomial ( logit )
Formula: Music_answer ~ 1 + (1 | exp_subject_id) + (1 | stim_name)
Data: Alldata
AIC BIC logLik -2*log(L) df.resid
11160.6 11184.4 -5577.3 11154.6 20157
Scaled residuals:
Min 1Q Median 3Q Max
-15.2054 -0.2813 -0.0531 0.1321 20.0969
Random effects:
Groups Name Variance Std.Dev.
exp_subject_id (Intercept) 1.93 1.389
stim_name (Intercept) 14.75 3.841
Number of obs: 20160, groups: exp_subject_id, 240; stim_name, 42
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.6018 0.5992 -1.004 0.315Effect_log <- glmer(data = Alldata, Music_answer ~ Order_letter + (1|exp_subject_id) + (1|stim_name), family = binomial)
summary(Effect_log)Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
Family: binomial ( logit )
Formula: Music_answer ~ Order_letter + (1 | exp_subject_id) + (1 | stim_name)
Data: Alldata
AIC BIC logLik -2*log(L) df.resid
11158.7 11190.3 -5575.3 11150.7 20156
Scaled residuals:
Min 1Q Median 3Q Max
-15.6036 -0.2817 -0.0522 0.1324 20.6263
Random effects:
Groups Name Variance Std.Dev.
exp_subject_id (Intercept) 1.932 1.390
stim_name (Intercept) 14.761 3.842
Number of obs: 20160, groups: exp_subject_id, 240; stim_name, 42
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.55223 0.60260 -0.916 0.3594
Order_letterretest -0.09978 0.04960 -2.012 0.0443 *
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Correlation of Fixed Effects:
(Intr)
Ordr_lttrrt -0.041anova(Effect_null,Effect_log)Data: Alldata
Models:
Effect_null: Music_answer ~ 1 + (1 | exp_subject_id) + (1 | stim_name)
Effect_log: Music_answer ~ Order_letter + (1 | exp_subject_id) + (1 | stim_name)
npar AIC BIC logLik -2*log(L) Chisq Df Pr(>Chisq)
Effect_null 3 11161 11184 -5577.3 11155
Effect_log 4 11159 11190 -5575.3 11151 3.9429 1 0.04707 *
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1Details about the full model
tidy (Effect_log)performance(Effect_log)# Indices of model performance
AIC | AICc | BIC | R2 (cond.) | R2 (marg.) | ICC | RMSE | Sigma | Log_loss | Score_log | Score_spherical
----------------------------------------------------------------------------------------------------------------------------
11158.680 | 11158.682 | 11190.325 | 0.835 | 1.245e-04 | 0.835 | 0.277 | 1.000 | 0.249 | -Inf | 5.847e-04ICC for the random effects (all together and separately)
# ICC calculation by group
icc_results_group <- performance::icc(Effect_log, by_group = TRUE)
print(icc_results_group)# ICC by Group
Group | ICC
----------------------
exp_subject_id | 0.097
stim_name | 0.739# ICC calculation for all random
icc_results_all <- performance::icc(Effect_log, by_group = FALSE)
print(icc_results_all)# Intraclass Correlation Coefficient
Adjusted ICC: 0.835
Unadjusted ICC: 0.835Figure 1C
Preparation for the figure with each retests (3 conditions) against the baseline (single for all “test”)
Alldata <- Alldata %>%
mutate(New_Group = case_when(
Order_letter == 'test' ~ 'Baseline',
Order_letter == 'retest' & Position == 'BB' ~ 'BB',
Order_letter == 'retest' & Position == 'WBR' ~ 'WBR',
Order_letter == 'retest' & Position == 'WBC' ~ 'WBC',
TRUE ~ 'Other'
))
Setting colors for the plots
baseline = "grey"
BB = "pink"
WBR = "cyan"
WBC = "magenta"
Position_colors = c(baseline,BB,WBR,WBC)Plotting the music answers with 3 colors (Conditions: Baseline, BB, WBC, WBR)
## From exp2:
# Compute mean music responses and mean confidence ratings
DataPlot <- with(Alldata, aggregate(Music_answer, list(Stimuli=stim_name,Condition=New_Group), mean))
DataPlot
Baseline_means <- with(subset(Alldata, New_Group == "Baseline"),
tapply(Music_answer, stim_name, mean, na.rm = TRUE))
# Reorder the Stimuli factor levels based on the means in the "self" condition
DataPlot$Stimuli <- factor(DataPlot$Stimuli,
levels = names(sort(Baseline_means, decreasing = FALSE)))
# Reorder the levels of the Condition factor
DataPlot$Condition <- factor(DataPlot$Condition, levels = c("Baseline", "BB", "WBR","WBC"))
# Create the plot with the reordered Condition levels
ResponsePlot <- ggplot(DataPlot, aes(Stimuli, y = x, group = Condition, colour = Condition)) +
geom_point(aes(shape = Condition), size = 3.5) +
scale_colour_manual(values = Position_colors,
labels = c("Baseline", "BB", "WBR","WBC")) +
scale_shape_manual(values = c("Baseline" = 16, # Solid circle
"BB" = 1,
"WBR"= 1,
"WBC"= 1), # Hollow circle
labels = c("Baseline", "BB", "WBR","WBC")) +
theme_classic() +
xlab("Stimuli (n = 42)") +
ylab("Music responses") +
theme(axis.text.x = element_blank(),
legend.text = element_text(size = 12),
legend.title = element_text(size = 12),
legend.position = c(0.25, 0.7), # Move legend to top-left
axis.title.y = element_text(size = 14),
axis.title.x = element_text(size = 12),
legend.spacing.y = unit(0.1, "cm")) + # Adjust space between legend items
theme(panel.background = element_rect(fill = "white"))
ResponsePlot
Create and export plot
Figure_1C <- grid.arrange(ResponsePlot,ncol = 1)
ggsave("~/Desktop/ResponsePlot_exp3a.tiff", plot = Figure_1C, width = 3, height = 2, dpi = 600, device = "tiff", compression = "lzw")Listeners’ consistence observed with test-retest phi-coefficients for each condition
consist_sub <- with(Alldata, aggregate(Music_answer, list(Sub=exp_subject_id, Stimuli=stim_name,Position=Position,Repetition=Order), mean))
consist_sub
consistency_sub <-reshape(consist_sub, idvar =c("Stimuli", "Position","Sub"), timevar = c("Repetition"), direction = "wide")
consistency_sub
sub_correlation <- consistency_sub %>%
group_by(Sub, Position) %>%
summarise(
phi_coefficient = {
tbl <- table(x.1, x.2)
assocstats(tbl)$phi
},
p_value = {
tbl <- table(x.1, x.2)
chisq.test(tbl)$p.value
}
)
print(sub_correlation)Plot listeners’ consistency between first and second presentation (grey dot for each stimulus on a x axis) for each Position separately (y axis)
# Plot the difference values for each group in 'Position'
# Function to calculate mean and confidence intervals
mean_ci <- function(x) {
mean_x <- mean(x)
ci <- qt(0.975, df = length(x) - 1) * sd(x) / sqrt(length(x))
return(data.frame(y = mean_x, ymin = mean_x - ci, ymax = mean_x + ci))
}
Position_colors_violin = c(BB,WBC,WBR)
# Plot itself
ConsistencyPlot <- ggplot(sub_correlation, aes(x = Position, y = phi_coefficient, colour = Position, fill = Position)) +
#geom_violin(alpha = 0.5) +
geom_boxplot(alpha = 0.5)+
scale_colour_manual(values = Position_colors_violin) +
scale_fill_manual(values = Position_colors_violin) +
geom_jitter(width = 0.2, alpha = 0.5) +
stat_summary(fun.data = mean_ci, geom = "pointrange", color = "black", size = 0.5) +
theme_classic() +
labs(#title = "Exp 3",
x = "Position",
y = "Phi coeficient") +
theme(legend.position="Position",
axis.title.y = element_text(size = 14),
axis.title.x = element_text(size = 12)) +
theme(panel.background = element_rect(fill = "white"))
ConsistencyPlot
Creation and export of figure 1C
Figure_1C <- grid.arrange(ResponsePlot,ConsistencyPlot,ncol = 1)
ggsave("~/Desktop/ResponsePlot_exp3.tiff", plot = Figure_1C, width = 3, height = 5, dpi = 600, device = "tiff", compression = "lzw")Comparison of listeners’ consistency (i.e., Phi coefficients) across conditions with anova and Turkey test
## Test the effect of Position on the Phi coefficients with (independent measures) Anova
consistency_anova <- aov(phi_coefficient~factor(Position), data = sub_correlation)
summary(consistency_anova) Df Sum Sq Mean Sq F value Pr(>F)
factor(Position) 2 0.9032 0.4516 40.46 7.65e-16 ***
Residuals 237 2.6454 0.0112
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1# Run the ANOVA
consistency_anova <- aov(phi_coefficient ~ factor(Position), data = sub_correlation)
summary(consistency_anova) Df Sum Sq Mean Sq F value Pr(>F)
factor(Position) 2 0.9032 0.4516 40.46 7.65e-16 ***
Residuals 237 2.6454 0.0112
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1# Perform Tukey HSD post hoc test
tukey_result <- TukeyHSD(consistency_anova)
# Display the Tukey HSD result
print(tukey_result) Tukey multiple comparisons of means
95% family-wise confidence level
Fit: aov(formula = phi_coefficient ~ factor(Position), data = sub_correlation)
$`factor(Position)`
diff lwr upr p adj
WBC-BB 0.1355419 0.09626492 0.17481894 0.0000000
WBR-BB 0.0117657 -0.02763617 0.05116758 0.7611985
WBR-WBC -0.1237762 -0.16329951 -0.08425295 0.0000000# Calculate mean and standard deviation of phi_coefficient per Position
summary_stats <- sub_correlation %>%
group_by(Position) %>%
summarise(
mean_phi = mean(phi_coefficient, na.rm = TRUE),
sd_phi = sd(phi_coefficient, na.rm = TRUE)
)
# Print the summary statistics
print(summary_stats)Plot of the two figures for Experiment 3
Figure_1C <- grid.arrange(ResponsePlot,ConsistencyPlot,ncol = 1)
EXPERIMENT 4 (Author: Talip Ata Aydin)
Loading the libraries
# Load everything else first
suppressWarnings(suppressMessages(library(readr)))
suppressWarnings(suppressMessages(library(tidyr)))
suppressWarnings(suppressMessages(library(ggplot2)))
suppressWarnings(suppressMessages(library(tidyverse)))
suppressWarnings(suppressMessages(library(psych)))
suppressWarnings(suppressMessages(library(readxl)))
suppressWarnings(suppressMessages(library(patchwork)))
suppressWarnings(suppressMessages(library(caTools)))
suppressWarnings(suppressMessages(library(car)))
suppressWarnings(suppressMessages(library(quantmod)))
suppressWarnings(suppressMessages(library(MASS)))
suppressWarnings(suppressMessages(library(corrplot)))
suppressWarnings(suppressMessages(library(effects)))
suppressWarnings(suppressMessages(library(writexl)))
suppressWarnings(suppressMessages(library(tidytable)))
suppressWarnings(suppressMessages(library(stringr)))
suppressWarnings(suppressMessages(library(ape)))
suppressWarnings(suppressMessages(library(grid)))
suppressWarnings(suppressMessages(library(gridExtra)))
suppressWarnings(suppressMessages(library(dplyr)))
Loading the data
demo_data <- read_excel("demo_data.xlsx",
sheet = "Sheet1")
Error: `path` does not exist: ‘demo_data.xlsx’MAIN TASK #############
We will create a dataframe for the main task, since the order of the sliders were randomized across 6 conditions, there are 6 different task names.
monra = music or not ratings (What do you think this is? 0 = not-music, 100 = music, all features are also rated on sliders from 0 to 100.)
ratings <- all_data %>%
dplyr::select(!monra) %>%
filter(task_name %in% c("Categorization", "Categorization2", "Categorization3", "Categorization4", "Categorization5", "Categorization6")) %>%
left_join(
all_data %>%
filter(task_name %in% c("Music_or_not")) %>%
dplyr::select(exp_subject_id, audio_name, monra),
join_by(exp_subject_id, audio_name)
) %>%
filter(!audio_group == "signal") %>%
dplyr::select(exp_subject_id, audio_name, monra, pulse, rhythm, repetition, melody, instrumental, timbre, tempo, intentionality, harmony, complexity, pag)
#Now the long format of the same dataframe
ratings_long <- pivot_longer(ratings,
cols = c(monra, pulse, rhythm, repetition, melody, instrumental, timbre, tempo, intentionality, harmony, complexity), # select columns to pivot
names_to = "variable", # new column name for variable names
values_to = "value") %>%
mutate(levels = case_when(
variable %in% c("instrumental", "intentionality", "complexity", "repetition") ~ 3,
variable %in% c("harmony", "melody", "rhythm") ~ 2,
variable %in% c("timbre", "pulse", "tempo") ~ 1,
TRUE ~ NA_integer_ # Default case if none of the conditions match
))
ratings_long_wo_monra <- ratings_long %>%
filter(!variable == "monra")
ratings_long_music <- ratings_long %>%
filter(variable == "monra")
ratings_long2 <- pivot_longer(ratings,
cols = c(pulse, rhythm, repetition, melody, instrumental, timbre, tempo, intentionality, harmony, complexity), # select columns to pivot
names_to = "variable", # new column name for variable names
values_to = "value") %>%
mutate(levels = case_when(
variable %in% c("instrumental", "intentionality", "complexity", "repetition") ~ 3,
variable %in% c("harmony", "melody", "rhythm") ~ 2,
variable %in% c("timbre", "pulse", "tempo") ~ 1,
TRUE ~ NA_integer_ # Default case if none of the conditions match
))Creating the dataframes for musical sophistication index (Gold-MSI) and auditory imagery (vividness, BAIS-V)
ratings_scales <- dplyr::select(all_data, exp_subject_id, task_name, audio_name, audio_group, monra, pulse, repetition, rhythm, melody, instrumental, timbre, tempo, intentionality, harmony, complexity, ae_01, ae_02, ae_05, ae_07, baisv_01, baisv_02, baisv_03, baisv_04, baisv_05, baisv_06, baisv_07, baisv_08, baisv_09, baisv_10, baisv_11, baisv_12, baisv_13, baisv_14, mt_01, mt_02, mt_03, mt_06, mt_07, pa_04, pa_08, sa_01, sa_02, sa_03, sa_04, sa_05, sa_06, em_04)
ratings <- all_data %>%
dplyr::select(exp_subject_id, task_name, audio_name, audio_group, monra, pulse, repetition, rhythm, melody, instrumental, timbre, tempo, intentionality, harmony, complexity) %>%
na.omit()
ratings <- all_data %>%
filter(task_name %in% c("Categorization", "Categorization2", "Categorization3", "Categorization4", "Categorization5", "Categorization6")) %>%
dplyr::select(exp_subject_id, audio_name, pulse, repetition, rhythm, melody, instrumental, timbre, tempo, intentionality, harmony, complexity) %>%
left_join(
all_data %>%
filter(!audio_group == "signal") %>%
filter(task_name %in% c("Music_or_not")) %>%
dplyr::select(exp_subject_id, audio_name, monra),
join_by(exp_subject_id, audio_name)
)
ratings <- merge(ratings, PAGG, by = "audio_name")
Filter only the categorization task and then filter out the signal sounds
#Only ratings of perceptual features
features_data <- ratings_scales %>%
filter(task_name %in% c("Categorization", "Categorization2", "Categorization3", "Categorization4", "Categorization5", "Categorization6"),
!audio_group == "signal") %>%
dplyr::select(exp_subject_id, audio_name, pulse, rhythm, repetition, melody, instrumental, timbre, tempo, intentionality, harmony, complexity)
#Only music or not ratings
monra_data <- ratings_scales %>%
filter(task_name %in% c("Music_or_not"),
!audio_group == "signal") %>%
dplyr::select(exp_subject_id, audio_name, monra)
write_xlsx(monra_data, "monra_data.xlsx")Error in write_xlsx(monra_data, "monra_data.xlsx") :
could not find function "write_xlsx"Data Frames for MM1 analysis
monra_mm1 <- monra_data %>%
pivot_wider(names_from = audio_name, values_from = monra)
pulse_mm1 <- features_data %>%
dplyr::select(exp_subject_id, audio_name, pulse) %>%
pivot_wider(names_from = audio_name, values_from = pulse)
repetition_mm1 <- features_data %>%
dplyr::select(exp_subject_id, audio_name, repetition) %>%
pivot_wider(names_from = audio_name, values_from = repetition)
rhythm_mm1 <- features_data %>%
dplyr::select(exp_subject_id, audio_name, rhythm) %>%
pivot_wider(names_from = audio_name, values_from = rhythm)
melody_mm1 <- features_data %>%
dplyr::select(exp_subject_id, audio_name, melody) %>%
pivot_wider(names_from = audio_name, values_from = melody)
instrumental_mm1 <- features_data %>%
dplyr::select(exp_subject_id, audio_name, instrumental) %>%
pivot_wider(names_from = audio_name, values_from = instrumental)
timbre_mm1 <- features_data %>%
dplyr::select(exp_subject_id, audio_name, timbre) %>%
pivot_wider(names_from = audio_name, values_from = timbre)
tempo_mm1 <- features_data %>%
dplyr::select(exp_subject_id, audio_name, tempo) %>%
pivot_wider(names_from = audio_name, values_from = tempo)
intentionality_mm1 <- features_data %>%
dplyr::select(exp_subject_id, audio_name, intentionality) %>%
pivot_wider(names_from = audio_name, values_from = intentionality)
harmony_mm1 <- features_data %>%
dplyr::select(exp_subject_id, audio_name, harmony) %>%
pivot_wider(names_from = audio_name, values_from = harmony)
complexity_mm1 <- features_data %>%
dplyr::select(exp_subject_id, audio_name, complexity) %>%
pivot_wider(names_from = audio_name, values_from = complexity)
A more descriptive DF that we will use, the mean ratings of all sliders (monra + features) at the STIMULUS level!
audio_descriptives <- ratings %>%
group_by(audio_name) %>%
summarize(
mean_monra = mean(monra, na.rm = TRUE),
mean_pulse = mean(pulse, na.rm = TRUE),
mean_repetition = mean(repetition, na.rm = TRUE),
mean_rhythm = mean(rhythm, na.rm = TRUE),
mean_melody = mean(melody, na.rm = TRUE),
mean_instrumental = mean(instrumental, na.rm = TRUE),
mean_timbre = mean(timbre, na.rm = TRUE),
mean_tempo = mean(tempo, na.rm = TRUE),
mean_intentionality = mean(intentionality, na.rm = TRUE),
mean_harmony = mean(harmony, na.rm = TRUE),
mean_complexity = mean(complexity, na.rm = TRUE))
audio_descriptives <- merge(audio_descriptives, PAGG, by = "audio_name")
audio_descriptives_long <- pivot_longer(audio_descriptives,
cols = c(mean_monra, mean_pulse, mean_rhythm, mean_repetition, mean_melody, mean_instrumental,
mean_timbre, mean_tempo, mean_intentionality, mean_harmony, mean_complexity),
names_to = "variable",
values_to = "value")We do the k-means clustering here. This is because the clusters will then be used for further analysis etc. and we thought it makes more sense to put the cluster analysis here.
Because R gives random numbers to clusters each time we run the analyses, we will simply upload the file with clusters, the code to find those clusters are commented out.
We decided that a 3-cluster-solution is the best. Looking at them, there is one with music stimuli (music cluster), another with not-music stimuli (not-music cluster, although in the code you might see it as “non-music”), and then a cluster in between, which we call the “ambiguous cluster”.
suppressWarnings(suppressMessages(library(readxl)))
# # Assuming you have loaded your data frame 'audio_descriptives' correctly
# justmeans <- dplyr::select(audio_descriptives, audio_name, mean_monra)
#
# # Perform k-means clustering
# kmm <- kmeans(justmeans$mean_monra, 3, nstart = 50, iter.max = 15)
# print(kmm)
# print(kmm$betweenss / kmm$totss)
#
# # Method 1: Elbow Method for finding the optimal number of clusters
# set.seed(123)
#
# # Compute and plot wss for k = 2 to k = 15
# k.max <- 15
# data <- justmeans$mean_monra # Ensure this is the correct column name
# data <- na.omit(data) # Remove NA values if any
#
# wss <- sapply(1:k.max, function(k) {
# kmeans(data, k, nstart = 50, iter.max = 15)$tot.withinss
# })
#
# # Plot the results
# plot(1:k.max, wss,
# type = "b", pch = 19, frame = FALSE,
# xlab = "Number of clusters",
# ylab = "Total Within-Clusters Sum of Squares")
#
#
#
# justmeans <- justmeans %>%
# mutate(cluster = kmm$cluster)
#
# # View the first few rows of the updated data frame
# head(justmeans)
#write_xlsx(justmeans, "justmeans.xlsx")
#Loading the data with stimuli, cluster labels, and cluster numbers (numbers will be irrelevant)
justmeans <- read_excel("justmeans.xlsx",
sheet = "Sheet1")
# Create a separate data frame called audio_clusters for clarity and future use
audio_clusters <- justmeans %>%
dplyr::select(audio_name, cluster) %>%
mutate(label = case_when(
cluster == 1 ~ "music",
cluster == 2 ~ "ambiguous",
cluster == 3 ~ "non-music"))
audio_clusters <- audio_clusters %>%
mutate(clean_names = audio_name %>%
gsub("^(M_|N_)", "", .) %>% # Remove M_ or N_ at the start
gsub("\\.mp3$", "", .) %>% # Remove .mp3 at the end
gsub("_", " ", .) %>% # Replace underscores with spaces
str_to_title() # Capitalize each word
)
audio_descriptives <- merge(audio_descriptives, audio_clusters, by = "audio_name")
#Including the cluster column to other dataframes
ratings <- ratings %>%
left_join(audio_clusters %>%
dplyr::select(audio_name, label, cluster),
by = "audio_name")
ratings_long <- ratings_long %>%
left_join(audio_clusters %>%
dplyr::select(audio_name, label),
by = "audio_name")
ratings_long_wo_monra <- ratings_long_wo_monra %>%
left_join(audio_clusters %>%
dplyr::select(audio_name, label),
by = "audio_name")
ratings_lme <- ratings
write_xlsx(audio_descriptives, "audio_descriptives.xlsx")
write_xlsx(ratings, "ratings.xlsx")
A dataframe for the means of all audios
audio_features <- dplyr::select(audio_descriptives, audio_name, mean_monra, mean_pulse, mean_rhythm, mean_repetition, mean_melody, mean_instrumental, mean_timbre, mean_tempo, mean_intentionality, mean_harmony, mean_complexity, label)
write_xlsx(audio_features, "audio_features.xlsx")
Participant means, average values at the participant level
participant_means <- ratings %>%
group_by(exp_subject_id) %>%
summarize(across(.cols = c(monra, pulse, rhythm, repetition, melody, instrumental, timbre, tempo, intentionality, harmony, complexity), .fns = mean))
Calculate Gold-MSI
gold <- dplyr::select(all_tasks, exp_subject_id, ae_01, ae_02, ae_05, ae_07, mt_01, mt_02, mt_03, mt_06, mt_07, pa_04, pa_08, sa_01, sa_02, sa_03, sa_04, sa_05, sa_06, em_04)
gold <- aggregated_data <- aggregate(. ~ exp_subject_id, data = gold, FUN = function(x) x[!is.na(x)][1])
gold2<-dplyr::select(gold, ! exp_subject_id)
keys_gold <- c("sa_05", "sa_04", "mt_03", "mt_07", "pa_08")
rec <- psych::reverse.code(keys = keys_gold, items = gold2, mini = 1, maxi = 7)
# View(rec)
rec <- as.data.frame(rec)
#GM_sum
rec$gold=rowSums(cbind(rec$ae_01,rec$ae_02,rec$ae_05,rec$ae_07,rec$em_04,rec$mt_01,rec$mt_02,rec$`mt_03-`,rec$mt_06,rec$`mt_07-`,rec$pa_04,rec$`pa_08-`,rec$sa_01,rec$sa_02,rec$sa_03,rec$`sa_04-`,rec$`sa_05-`,rec$sa_06))
rec$exp_subject_id <- gold$exp_subject_id
gold_score <- dplyr::select(rec, exp_subject_id, gold)
Calculate BAIS-V
bais <- dplyr::select(all_tasks, exp_subject_id, baisv_01, baisv_02, baisv_03, baisv_04, baisv_05, baisv_06, baisv_07, baisv_08, baisv_09, baisv_10, baisv_10, baisv_11, baisv_12, baisv_13, baisv_14)
bais <- na.omit(bais)
rec <- as.data.frame(bais)
#Mean_score
rec$bais=rowMeans(cbind(rec$baisv_01,rec$baisv_02,rec$baisv_03,rec$baisv_04,rec$baisv_05,rec$baisv_06,rec$baisv_07,rec$baisv_08,rec$baisv_09,rec$baisv_10,rec$baisv_11,rec$baisv_12,rec$baisv_13,rec$baisv_14))
BAIS_score<-dplyr::select(rec,bais)
BAIS_score <- cbind(BAIS_score, bais$exp_subject_id)
colnames(BAIS_score)[colnames(BAIS_score) == "BAIS_score$exp_subject_id"] <- "exp_subject_id"
scales <-cbind(gold_score, BAIS_score)
Calculate Openness
openness <- dplyr::select(all_tasks, exp_subject_id, open1, open2, open3, open4, open5, open6, open7, open8, open9, open10)
reverse_code <- function(x) {
return(6 - x)
}
# Apply reverse coding and calculate total_openness
openness <- openness %>%
mutate(
open2 = reverse_code(open2),
open4 = reverse_code(open4),
open6 = reverse_code(open6),
open8 = reverse_code(open8),
open10 = reverse_code(open10)
) %>%
group_by(exp_subject_id) %>%
summarise(total_openness = sum(c_across(open1:open10), na.rm = TRUE))
# View the result
print(openness)# A tidytable: 98 × 2
exp_subject_id total_openness
<dbl> <dbl>
1 937114 39
2 937115 37
3 937117 46
4 937126 44
5 937138 43
6 937151 39
7 937164 36
8 937267 42
9 943497 36
10 943525 29
# ℹ 88 more rows
# ℹ Use `print(n = ...)` to see more rows# Merge all scales in one dataframe
all_scales <- merge(scales, openness, by ="exp_subject_id")
all_scales <- all_scales
mean(all_scales$total_openness)[1] 38.18367sd(all_scales$total_openness)[1] 6.381912min(all_scales$total_openness)[1] 18max(all_scales$total_openness)[1] 50mean(all_scales$bais)[1] 4.786443sd(all_scales$bais)[1] 1.020539min(all_scales$bais)[1] 1max(all_scales$bais)[1] 6.928571mean(all_scales$gold)[1] 71.45918sd(all_scales$gold)[1] 19.78397min(all_scales$gold)[1] 29max(all_scales$gold)[1] 114Let’s compute how well participants understood the definitions of features that we provided!
understanding <- all_tasks %>%
dplyr::select(exp_subject_id, pulseund, melodyund, instrumentalund, timbreun, tempound, intentionalityund, harmonyund, complexityund, toneund, rhythmund) %>%
na.omit() %>%
group_by(exp_subject_id) %>%
summarise(total_und = sum(c_across(pulseund:complexityund)))
understanding_perfeature <- all_tasks %>%
dplyr::select(exp_subject_id, pulseund, melodyund, instrumentalund, timbreun, tempound, intentionalityund, harmonyund, complexityund, toneund, rhythmund) %>%
na.omit()
understanding_long <- pivot_longer(understanding_perfeature,
cols = c(pulseund, melodyund, instrumentalund, timbreun,
tempound, intentionalityund, harmonyund, complexityund, toneund, rhythmund),
names_to = "variable",
values_to = "value")
understanding <- merge(understanding, all_scales, by ="exp_subject_id")
Let’s compute how often participants listen to different genres
genres <- all_tasks %>%
filter(task_name %in% c("Genres", "Genres2", "Genres3", "Genres4", "Genres5", "Genre6")) %>%
dplyr::select(exp_subject_id, top40, pop, rock, classical, noise, experimental, edm)
genres[is.na(genres)] <- 0
genres <- genres %>%
mutate(music_diversity = noise + experimental,
noise_experimental_listen = ifelse(is.na(music_diversity), "no", "yes"))
genres$noise_experimental_listen <- as.factor(genres$noise_experimental_listen)
all_scales <- all_scales %>%
left_join(genres %>%
dplyr::select(exp_subject_id, music_diversity),
by = "exp_subject_id")
Calculate the mean age and gender information
mean(demo_data$age)[1] 34.97959sd(demo_data$age)[1] 11.32098min(demo_data$age)[1] 18max(demo_data$age)[1] 74table(demo_data$gender)
female male
30 68 Put the average rating for each participant to the scales df
means_scales <- merge(all_scales, participant_means, by = "exp_subject_id")
means_scales_age <- means_scales %>%
left_join(demo_data %>%
dplyr::select(exp_subject_id, age),
by = "exp_subject_id")
means_scales_age_genres <- merge(means_scales_age, genres, by = "exp_subject_id")
A DF just for the ambiguous stimuli
ambiguous_stimuli <- ratings %>%
filter(label == "ambiguous")
ambiguous_participant_means <- ambiguous_stimuli %>%
group_by(exp_subject_id) %>%
summarize(mean_monra = mean(monra)) %>%
dplyr::select(exp_subject_id, mean_monra)
A DF for the descriptive stats for monra
monra_desc_stats <- ratings |>
group_by(audio_name) |>
summarize(mean_monra = mean(monra),
sd_monra = sd(monra)) |>
left_join(audio_clusters |>
dplyr::select(audio_name, label),
by = "audio_name")
monra_desc_stats$label <- factor(monra_desc_stats$label, levels = c("non-music", "ambiguous", "music"))
Circular cluster plot for the paper, included in Figure 2A
library(ape)
library(grid)
library(gridExtra)
# Reorder the audio_name based on the label to group the clusters
audio_clusters <- audio_clusters[order(audio_clusters$label), ]
audio_clusters$audio_name <- as.character(audio_clusters$audio_name)
# Step 1: Order the dataframe alphabetically including prefixes
audio_clusters <- audio_clusters %>%
arrange(audio_name)
audio_clusters$number <- 1:nrow(audio_clusters) # Assign numbers sequentially
audio_clusters <- audio_clusters %>%
mutate(clean_names = str_replace(audio_name, '^(.+)\\.mp3$', '\\1')) %>%
mutate(clean_names = str_to_title(clean_names))
distance_mat <- dist(audio_descriptives$mean_monra, method = 'euclidean')
distance_mat 1 2 3 4 5 6 7 8 9
2 9.93877551
3 11.68367347 1.74489796
4 12.50000000 22.43877551 24.18367347
5 9.28571429 19.22448980 20.96938776 3.21428571
6 2.65306122 12.59183673 14.33673469 9.84693878 6.63265306
7 3.62244898 6.31632653 8.06122449 16.12244898 12.90816327 6.27551020
8 1.95918367 11.89795918 13.64285714 10.54081633 7.32653061 0.69387755 5.58163265
9 10.42857143 0.48979592 1.25510204 22.92857143 19.71428571 13.08163265 6.80612245 12.38775510
10 12.64285714 2.70408163 0.95918367 25.14285714 21.92857143 15.29591837 9.02040816 14.60204082 2.21428571
11 21.57142857 31.51020408 33.25510204 9.07142857 12.28571429 18.91836735 25.19387755 19.61224490 32.00000000
12 12.53061224 2.59183673 0.84693878 25.03061224 21.81632653 15.18367347 8.90816327 14.48979592 2.10204082
10 11 12 13 14 15 16 17 18
2
3
4
5
6
7
8
9
10
11 34.21428571
12 0.11224490 34.10204082
19 20 21 22 23 24 25 26 27
2
3
4
5
6
7
8
9
10
11
12
28 29 30 31 32 33 34 35 36
2
3
4
5
6
7
8
9
10
11
12
37 38 39 40 41 42 43 44 45
2
3
4
5
6
7
8
9
10
11
12
46 47 48 49 50 51 52 53 54
2
3
4
5
6
7
8
9
10
11
12
55 56 57 58 59 60 61 62 63
2
3
4
5
6
7
8
9
10
11
12
64 65 66 67 68 69 70 71 72
2
3
4
5
6
7
8
9
10
11
12
73 74 75 76 77 78 79 80 81
2
3
4
5
6
7
8
9
10
11
12
82 83 84 85 86 87 88 89
2
3
4
5
6
7
8
9
10
11
12
[ reached 'max' / getOption("max.print") -- omitted 78 rows ]# Fitting Hierarchical clustering Model to training dataset
set.seed(240) # Setting seed
Hierar_cl <- hclust(distance_mat, method = "average")
# Assuming 'Hierar_cl' is your hclust result and 'fit' contains cluster assignments
phylo_tree <- as.phylo(Hierar_cl) # Convert hclust to phylo object
fit <- cutree(Hierar_cl, k = 3)
# Define your cluster colors
# Ensure 'fit' is a vector containing cluster assignments for each tip
colors <- c("red", "blue", "green") # 1 = red, 2 = blue, 3 = green
tip_colors <- colors[fit] # Map the cluster assignments to colors
# Plot the circular dendrogram with color-coded tips
plot(phylo_tree,
type = "fan", # Circular plot
tip.color = tip_colors, # Color tips based on clusters
label.offset = 0.01, # Space between tips and labels
cex = 1.6, # Text size
show.tip.label = TRUE) # Show labels (audio names)
# Define colors for each cluster
label_colors <- c("music" = "red", "ambiguous" = "blue", "non-music" = "green")
# Create a blank plot for printing the text
grid.newpage()
# Set up the column positions
x_positions <- c(0.1, 0.4, 0.7) # X coordinates for the 3 columns
# Loop through each row and assign text to the correct column and row position
for (i in 1:nrow(audio_clusters)) {
# Determine the column (1st, 2nd, or 3rd column)
column <- ((i - 1) %/% 30) + 1
row <- (i - 1) %% 30 + 1
# Calculate the y position for each row (invert row index for grid positioning)
y_position <- 1 - (row * 0.03) # Adjust spacing between rows
# Print the text in the correct column, color-coded by cluster
grid.text(
label = paste(audio_clusters$number[i], "-", audio_clusters$clean_names[i]),
x = x_positions[column],
y = y_position,
gp = gpar(col = label_colors[audio_clusters$label[i]], fontsize = 20),
just = "left"
)
}
NA
NA
NA
NA
NAPlotting the participants and audios ordered by mean (Figure 2B)
The first column represent the mean music ratings ordered from high (on top) to low (on the bottom). Then, every other column is the ranking of each participant’s music ratings.
#Do this otherwise it overrides dplyr
detach("package:tidytable", unload = TRUE)
# Reshaping the dataframe
monra_mm1_long <- monra_mm1 %>%
pivot_longer(
cols = -exp_subject_id, # Keep exp_subject_id, gather other columns
names_to = "audio_name", # The names of the audio columns will go to this new column
values_to = "rating" # The ratings will go to this new column
)
# Now pivot wider so that exp_subject_id becomes columns
monra_mm1_wide <- monra_mm1_long %>%
pivot_wider(
names_from = exp_subject_id, # The exp_subject_id values become new columns
values_from = rating # Values come from the rating column
)
# Print the reshaped dataframe
print(monra_mm1_wide)
monra_mean_participants <- monra_mm1_wide %>%
left_join(audio_descriptives %>% dplyr::select(audio_name, mean_monra), by = "audio_name") %>%
relocate(mean_monra, .after = audio_name) %>%
left_join(audio_clusters %>% dplyr::select(audio_name, label), by = "audio_name") %>%
relocate(label, .after = mean_monra)
# Step 1: Reshape the original data to long format for participant ratings
long_data <- monra_mean_participants %>%
pivot_longer(
cols = -c(audio_name, mean_monra, label), # Exclude non-participant columns
names_to = "participant_id",
values_to = "rating"
) %>%
group_by(participant_id) %>%
arrange(rating, .by_group = TRUE) %>%
mutate(rank = row_number()) # Assign rank based on the order of ratings within each participant
# Step 2: Create a long format for mean_monra
mean_monra_long <- monra_mean_participants %>%
dplyr::select(audio_name, mean_monra, label) %>%
mutate(participant_id = "mean_monra", rating = mean_monra) %>%
dplyr::select(audio_name, participant_id, rating, label)
# Step 3: Combine both long formats
combined_long_data <- bind_rows(long_data, mean_monra_long)
# Step 4: Adjust the participant_id to make mean_monra first
combined_long_data <- combined_long_data %>%
mutate(participant_id = factor(participant_id, levels = c("mean_monra", unique(long_data$participant_id)))) # Ensures mean_monra is first
# Step 5: Rank the combined data
combined_long_data <- combined_long_data %>%
group_by(participant_id) %>%
arrange(rating, .by_group = TRUE) %>%
mutate(rank = row_number()) # Assign rank for all participants including mean_monra
# Step 6: Create the plot
ggplot(combined_long_data, aes(x = participant_id, y = rank, fill = label)) +
geom_tile() +
scale_fill_manual(
values = c("music" = "red", "ambiguous" = "blue", "non-music" = "green"),
labels = c("music" = "Music", "ambiguous" = "Ambiguous", "non-music" = "Not-music")
) +
labs(
x = "Participants",
y = "Ranking of Stimuli",
title = " ",
fill = "Cluster"
) +
theme_minimal() +
theme(
axis.text.x = element_blank(),
axis.title.x = element_text(size = 30),
axis.text.y = element_blank(),
axis.title.y = element_text(size = 30),
panel.grid = element_blank(),
legend.text = element_text(size = 15),
legend.title = element_text(size = 20),
axis.text.x.left = element_text(
ifelse(unique(combined_long_data$participant_id) == "mean_monra", "black", "transparent")
)
)
NA
NAExperiment 4 Analyses
monra_mean_participants %>%
pivot_longer(
cols = -c(audio_name, mean_monra, label),
names_to = "participant_id",
values_to = "rating"
) %>%
table(sapply(long_data$rating, class))```
Error: attempt to use zero-length variable nameaudio_descriptives <- read_excel("audio_descriptives.xlsx",
sheet = "Sheet1")
monra_data <- read_excel("monra_data.xlsx",
sheet = "Sheet1")
MM1 for monra
Overall MM1 value for music ratings is .87. This is very high, so participants to agree with each other. This is reported in the paper.
MM1_revised <- function(df){
#save & remove subj id
id <- df[, 1]
df <- as.matrix(df[,-1])
#Compute means minus one
means <- matrix(0, nrow = nrow(df), ncol = ncol(df))
for(j in 1:nrow(df)){
means[j,] <- colMeans(df[-j,])
}
#Compute correlations between means minus ones and ones ratings
mm1 <- sapply(1:nrow(df), function(.x) cor(df[.x,], means[.x,]))
#transform correlations into z scores (r to z Fisher)
z_scores <- sapply(mm1, function(x) log((1 + x)/(1 - x))) * .5
#save average correlation (z average to r average Fisher) and correlations per rater
result <- list(MM1 = (exp(2 * (mean(z_scores))) - 1)/(exp( 2 * (mean(z_scores))) + 1),
raw = data.frame(id, mm1 = mm1, mm1_z = z_scores))
result
}
#I wanted to see the MM1 list pls
MM1_List <- MM1_revised(monra_mm1)
MM1_L <- as.data.frame(MM1_List$raw)
MM1_List$MM1
[1] 0.8693654
$rawMM1_List$MM1[1] 0.8693654MM1 for each features (Reported)
This varies… For higher level features, there is more agreement, for lower level features, there is less…
MM1_repetition <- MM1_revised(repetition_mm1)
MM1_repetition$MM1[1] 0.4167283MM1_pulse <- MM1_revised(pulse_mm1)
MM1_pulse$MM1[1] 0.6281115MM1_rhythm <- MM1_revised(rhythm_mm1)
MM1_rhythm$MM1[1] 0.6734365MM1_melody <- MM1_revised(melody_mm1)
MM1_melody$MM1[1] 0.787064MM1_instrumental <- MM1_revised(instrumental_mm1)
MM1_instrumental$MM1[1] 0.8323992MM1_timbre <- MM1_revised(timbre_mm1)
MM1_timbre$MM1[1] 0.6026743MM1_tempo <- MM1_revised(tempo_mm1)
MM1_tempo$MM1[1] 0.5776043MM1_intentionality <- MM1_revised(intentionality_mm1)
MM1_intentionality$MM1[1] 0.7481585MM1_harmony <- MM1_revised(harmony_mm1)
MM1_harmony$MM1[1] 0.7194582MM1_complexity <- MM1_revised(complexity_mm1)
MM1_complexity$MM1[1] 0.6734471Krippendorff’s alpha for each rating, at the group level. (Reported)
monra_krip <- monra_mm1 %>%
dplyr::select(!exp_subject_id)
monra_krip <- as.matrix(monra_krip)
kripp.alpha(monra_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.684 intentionality_krip <- intentionality_mm1 %>%
dplyr::select(!exp_subject_id)
intentionality_krip <- as.matrix(intentionality_krip)
kripp.alpha(intentionality_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.478 instrumental_krip <- instrumental_mm1 %>%
dplyr::select(!exp_subject_id)
instrumental_krip <- as.matrix(instrumental_krip)
kripp.alpha(instrumental_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.622 complexity_krip <- complexity_mm1 %>%
dplyr::select(!exp_subject_id)
complexity_krip <- as.matrix(complexity_krip)
kripp.alpha(complexity_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.367 repetition_krip <- repetition_mm1 %>%
dplyr::select(!exp_subject_id)
repetition_krip <- as.matrix(repetition_krip)
kripp.alpha(repetition_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.115 melody_krip <- melody_mm1 %>%
dplyr::select(!exp_subject_id)
melody_krip <- as.matrix(melody_krip)
kripp.alpha(melody_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.539 rhythm_krip <- rhythm_mm1 %>%
dplyr::select(!exp_subject_id)
rhythm_krip <- as.matrix(rhythm_krip)
kripp.alpha(rhythm_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.369 harmony_krip <- harmony_mm1 %>%
dplyr::select(!exp_subject_id)
harmony_krip <- as.matrix(harmony_krip)
kripp.alpha(harmony_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.422 timbre_krip <- timbre_mm1 %>%
dplyr::select(!exp_subject_id)
timbre_krip <- as.matrix(timbre_krip)
kripp.alpha(timbre_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.285 tempo_krip <- tempo_mm1 %>%
dplyr::select(!exp_subject_id)
tempo_krip <- as.matrix(tempo_krip)
kripp.alpha(tempo_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.247 pulse_krip <- pulse_mm1 %>%
dplyr::select(!exp_subject_id)
pulse_krip <- as.matrix(pulse_krip)
kripp.alpha(pulse_krip, method = "interval") Krippendorff's alpha
Subjects = 90
Raters = 98
alpha = 0.31 Linear Mixed Effects Model to predict MoNRa by perceptual features (reported)
ACOUSTIC FEATURE SELECTION
Get and prepare data:
options(scipen=999)
#The features have been extracted already, we read in the data
acoustics_mir<- read_excel("essentia_acoustics_2.xlsx",
sheet = "Sheet1")
# We clean the names
acoustics_mir$audio_name <- sub("\\.mp3.*", ".mp3", acoustics_mir$audio_name)
# Get the column names
col_names <- colnames(acoustics_mir)
col_names
# Remove "" from the end of each column name
clean_col_names <- gsub('""$', '', col_names)
# Assign the cleaned column names back to the dataframe
colnames(acoustics_mir) <- clean_col_names
# Check the cleaned column names
print(colnames(acoustics_mir))
acoustics_mir <- acoustics_mir %>%
left_join(audio_descriptives %>%
dplyr::select(audio_name, label, mean_monra),
by = "audio_name") %>%
dplyr::select(audio_name, label, mean_monra, everything())
acoustics_mir <- acoustics_mir %>%
dplyr::select(!id) %>%
dplyr::select(!average_loudness.value)
acoustics_mir <- acoustics_mir %>% dplyr::select(, 1:251)
names(acoustics_mir)
Scale data:
scaled_mir_ <- acoustics_mir |>
select(!mean_monra) |>
mutate_if(is.numeric,scale)
# Scale columns 4 to 251
scaled_columns <- scale(acoustics_mir[, 4:251])
# Combine the unscaled first three columns with the scaled columns
acoustics_mir_scaled <- cbind(acoustics_mir[, 1:3], as.data.frame(scaled_columns))
#rownames(numeric_mir) <- scaled_mir_[,1]
str(numeric_mir)
names(scaled_mir_)
names(acoustics_mir)
First clean out obviously redundant features. barkbands.dmean_01-27 are the same as frequency_bands.dmean_01-27; subband_mean/std (2-10) same as mfcc2_mean/std; also remove spectral_energyband_low.dvar
These are 72 acoustic features that do not have correlations above .75 with each other.
highly_corr<-fuzzySim::corSelect(acoustics_mir_scaled[, 4:251], cor.thresh = .75, var.cols =1:248)
redundant_variables<-highly_corr$excluded.vars[1:176]
count(highly_corr$high.correlations)
numeric_mir_excluded <- numeric_mir %>% dplyr::select(!all_of(redundant_variables))
acoustics_mir_scaled_nocollinear <- acoustics_mir_scaled %>% dplyr::select(!all_of(redundant_variables))
names(numeric_mir)Now, we correlate each acoustic feature with mean music ratings to see how the relationships look like
acoustic_features_to_correlate <- acoustics_mir_scaled_nocollinear[,
sapply(acoustics_mir_scaled_nocollinear, is.numeric) &
names(acoustics_mir_scaled_nocollinear) != "mean_monra"
]
# Compute correlations
full_cor_results <- sapply(acoustic_features_to_correlate, function(column) {
cor(column, acoustics_mir_scaled_nocollinear$mean_monra, use = "complete.obs")
})
# Format results
full_cor_df <- data.frame(
Variable = names(full_cor_results),
Correlation = full_cor_results,
stringsAsFactors = FALSE
)
# Order by absolute correlation
full_cor_df <- full_cor_df[order(-abs(full_cor_df$Correlation)), ]
# Show result
print(full_cor_df)
Now, select the 10 acoustic features that have the abolsute highest correlation to mean_monra
acoustics_cluster_absolute_corr <- dplyr::select(acoustics_mir_scaled_nocollinear, audio_name, label, mean_monra, pitch_salience.dmean, silence_rate_30dB.dmean, scvalleys.dvar_01, gfcc.dmean_11, gfcc.dmean_09, zerocrossingrate.dvar, pitch_instantaneous_confidence.dvar, spectral_contrast.dvar_01, barkbands.dmean_23, gfcc.dmean_13)
write_xlsx(acoustics_cluster_absolute_corr, "acoustics_cluster_absolute_corr.xlsx")
write_xlsx(acoustics_mir_scaled_nocollinear, "acoustics_mir_scaled_nocollinear.xlsx")Comparing acoustic vs. perceptual features in predicting a music category
Packages
suppressWarnings(suppressMessages(library(broom.mixed)))
suppressWarnings(suppressMessages(library(cowplot)))
suppressWarnings(suppressMessages(library(factoextra)))
suppressWarnings(suppressMessages(library(ggfortify)))
suppressWarnings(suppressMessages(library(lme4)))
suppressWarnings(suppressMessages(library(lmerTest)))
suppressWarnings(suppressMessages(library(ordinal)))
suppressWarnings(suppressMessages(library(rcompanion)))
suppressWarnings(suppressMessages(library(stats)))
suppressWarnings(suppressMessages(library(writexl)))
Reading in DFs
acoustics_mir_scaled_nocollinear <- read_excel("acoustics_mir_scaled_nocollinear.xlsx",
sheet = "Sheet1")
audio_descriptives <- read_excel("audio_features.xlsx",
sheet = "Sheet1")
monra_data <- read_excel("monra_data.xlsx",
sheet = "Sheet1")
acoustics_cluster_absolute_corr <- read_excel("acoustics_cluster_absolute_corr.xlsx",
sheet = "Sheet1")
PCA WITH PERCEPTUAL FEATURES
We reduce the dimensionality of our perceptual features to later use these principal components in modeling. We select the first two principal components here.
pca_result_perceptual <- prcomp(audio_descriptives[,3:12], center = TRUE, scale. = TRUE)
fviz_eig(pca_result_perceptual, addlabels = TRUE)
eig.val_per <- get_eigenvalue(pca_result_perceptual)
eig.val_per
# number of PCs to keep
pcdims_per = 10
# Print eigenvalues
eig.val_per[c(1:pcdims_per),]
res.feats_per <- get_pca_var(pca_result_perceptual)
contribdf_per = as.data.frame(res.feats_per$contrib[,c(1:pcdims_per)])
round(contribdf_per, digits=2)
loadings_per <- pca_result_perceptual$rotation
loadings_per[,c(1:2)]
res.ind_per <- get_pca_ind(pca_result_perceptual)
pc_scores_per = pca_result_perceptual$x
pca_data_perc <- as.data.frame(pca_result_perceptual$x)
pca_data_perc$label <- mir_monra2$label
audio_descriptives$PC1 = pc_scores_per[,1]*-1
audio_descriptives$PC2 = pc_scores_per[,2]
# Now re-run the plotting
fviz_pca_biplot(pca_result_perceptual,
label = "var", # only variable names
habillage = mir_monra2$label,
addEllipses = TRUE,
col.var = "black",
repel = TRUE)
fviz_eig(pca_result_perceptual, addlabels = TRUE, ylim = c(0, 100)) +
theme_minimal() + # Minimal theme for a clean look
theme(
panel.grid.major = element_blank(), # Remove major grid lines
panel.grid.minor = element_blank(), # Remove minor grid lines
panel.background = element_blank(), # Remove background color
axis.line = element_line(colour = "black"), # Add axis lines
legend.position = "NONE", # Remove the legend
text = element_text(size = 16))
PCA to reduce the number of acoustic features, we select the first 2 principal components.
acoustics_cluster_absolute_corr <- acoustics_cluster_absolute_corr %>%
mutate(order = ifelse(label == "non-music", 1,
ifelse(label == "ambiguous", 2,
ifelse(label == "music", 3, NA))))
acoustics_cluster_absolute_corr$order <- as.factor(acoustics_cluster_absolute_corr$order)
pca_result_abs <- prcomp(acoustics_cluster_absolute_corr[4:13], center = TRUE, scale. = FALSE)
fviz_eig(pca_result_abs, addlabels = TRUE, ylim = c(0, 100)) +
theme_minimal() + # Minimal theme for a clean look
theme(
panel.grid.major = element_blank(), # Remove major grid lines
panel.grid.minor = element_blank(), # Remove minor grid lines
panel.background = element_blank(), # Remove background color
axis.line = element_line(colour = "black"), # Add axis lines
legend.position = "NONE", # Remove the legend
text = element_text(size = 16))
# Plot the first 2 principal components
autoplot(pca_result_abs,
data = acoustics_cluster_absolute_corr,
x = 1,
y = 2,
colour = 'label',
loadings = TRUE,
loadings.label = TRUE,
loadings.color = "black",
loadings.label.color = "black",
loadings.label.repel = TRUE,
loadings.label.size = 5,
frame = TRUE,
frame.type = 'norm') +
# Customize appearance
theme_minimal() +
theme(
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.line = element_line(colour = "black"),
legend.position = "NONE",
text = element_text(size = 30)
) +
scale_color_manual(values = c("music" = "red", "ambiguous" = "blue", "non-music" = "green")) +
scale_fill_manual(values = c("music" = "red", "ambiguous" = "blue", "non-music" = "green")) +
labs(
title = " ",
colour = "Mean MONRA Rating"
) +
scale_y_reverse()
pc_scoreabs = pca_result_abs$x
acoustics_cluster_absolute_corr$PC1ac = pc_scoreabs[,1]
acoustics_cluster_absolute_corr$PC2ac = pc_scoreabs[,2]Here, we carry out 2 cumulative link models to predict categories using the ordinal package in 2 separate models: acoustic model and the perceptual model
clm_df <- audio_descriptives |>
dplyr::select(audio_name, PC1, PC2) |>
left_join(acoustics_cluster_absolute_corr |>
dplyr::select(audio_name, PC1ac, PC2ac),
by = "audio_name") |>
left_join(ratings_order |>
distinct(audio_name, order),
by = "audio_name")
clm_ac_abs <- clm(order ~ PC1ac + PC2ac, data = clm_df)
clm_perceptual <- clm(order ~ PC1 + PC2, data = clm_df)
clm_order_null <- clm(order ~ 1, data = clm_df)
summary(clm_ac_abs)
summary(clm_perceptual)
anova(clm_ac_abs, clm_order_null)
anova(clm_perceptual, clm_order_null)
pR2(lm_order_ac)
###Pseudo r sqaured
nagelkerke(clm_ac_abs, clm_order_null)
nagelkerke(clm_perceptual, clm_order_null)
# Compare AIC
AIC(clm_perceptual, clm_ac_abs)
# Compare BIC
BIC(clm_perceptual, clm_ac_abs)
write_xlsx(clm_df, "clm_df.xlsx")Hence, the perceptual model outperforms the acoutic model.
Reported PCA Plots (Figure 3A)
library(cowplot)
acoustic_pca_plot <- autoplot(pca_result_abs,
data = acoustics_cluster_absolute_corr,
colour = 'label',
loadings = TRUE,
loadings.label = TRUE,
loadings.color = "black",
loadings.label.color = "black",
loadings.label.repel = TRUE,
loadings.label.size = 5,
frame = TRUE,
frame.type = 'norm') +
# Customize appearance
theme_minimal() + # Minimal theme for a clean look
theme(
panel.grid.major = element_blank(), # Remove major grid lines
panel.grid.minor = element_blank(), # Remove minor grid lines
panel.background = element_blank(), # Remove background color
axis.line = element_line(colour = "black"), # Add axis lines
legend.position = "NONE", # Remove the legend
text = element_text(size = 30) # Increase text size
) +
scale_color_manual(values = c("music" = "red", "ambiguous" = "blue", "non-music" = "green")) +
scale_fill_manual(values = c("music" = "red", "ambiguous" = "blue", "non-music" = "green")) + # Custom colors
labs(
title = " ",
colour = "Mean MONRA Rating" # Label for the legend
)
perceptual_pca_plot <- autoplot(pca_result_perceptual,
data = audio_descriptives,
colour = 'label',
loadings = TRUE,
loadings.label = TRUE,
loadings.color = "black",
loadings.label.color = "black",
loadings.label.repel = TRUE,
loadings.label.size = 5,
frame = TRUE,
frame.type = 'norm') +
# Customize appearance
theme_minimal() + # Minimal theme for a clean look
theme(
panel.grid.major = element_blank(), # Remove major grid lines
panel.grid.minor = element_blank(), # Remove minor grid lines
panel.background = element_blank(), # Remove background color
axis.line = element_line(colour = "black"), # Add axis lines
legend.position = "NONE", # Remove the legend
text = element_text(size = 30) # Increase text size
) +
scale_color_manual(values = c("music" = "red", "ambiguous" = "blue", "non-music" = "green")) +
scale_fill_manual(values = c("music" = "red", "ambiguous" = "blue", "non-music" = "green")) + # Custom colors
labs(
title = " ",
colour = "Mean MONRA Rating" # Label for the legend
)
legend_pca <- ggplot(data = audio_descriptives, aes(x = 1, y = 1, colour = label)) +
geom_point() +
scale_color_manual(values = c("music" = "red", "ambiguous" = "blue", "non-music" = "green"),
labels = c("Ambiguous", "Music", "Non-music")) +
theme_void() +
theme(legend.position = "right") +
guides(colour = guide_legend(override.aes = list(size = 5))) +
labs(colour = NULL) # Remove the legend title
loadings_labels <- rownames(pca_result_perceptual$rotation)
# Remove "mean_" from the loadings labels and capitalize the first letter
loadings_labels <- gsub("^mean_", "", loadings_labels) # Remove "mean_"
loadings_labels <- tools::toTitleCase(loadings_labels) # Capitalize first letter
# Reassign the modified labels back to the PCA object
rownames(pca_result_perceptual$rotation) <- loadings_labels
cowplot::plot_grid(acoustic_pca_plot, perceptual_pca_plot, legend_pca, ncol = 3, rel_widths = c(0.8, 0.8, 0.1))
Boxplots of mean feature ratings across clusters (Figure 3B)
audio_features_long <- pivot_longer(audio_features,
cols = c(mean_pulse, mean_rhythm, mean_repetition, mean_melody, mean_instrumental,
mean_timbre, mean_tempo, mean_intentionality, mean_harmony, mean_complexity),
names_to = "variable",
values_to = "value")
audio_features_long$variable <- as.factor(audio_features_long$variable)
audio_features_long$label <- factor(audio_features_long$label, levels = c("music", "ambiguous", "non-music"))
# Modify the 'variable' column: Remove "mean_" and capitalize the first letter
audio_features_long <- audio_features_long %>%
mutate(variable = gsub("^mean_", "", variable), # Remove "mean_"
variable = tools::toTitleCase(variable)) # Capitalize first letter
audio_features_long$variable <- factor(audio_features_long$variable,
levels = c("Intentionality", "Instrumental", "Complexity", "Repetition", "Melody", "Harmony", "Rhythm", "Tempo", "Timbre", "Pulse"))
ratings_cluster_plot_mean <- ggplot(audio_features_long, aes(x = label, y = value, color = label, fill = label)) +
geom_boxplot(alpha = 0.5, position = position_dodge(width = 0.75)) +
geom_jitter(alpha = 0.3, size = 1) +
theme_minimal(base_size = 16) + # Increase base font size
theme(
axis.title = element_text(size = 30),
plot.title = element_text(size = 18, face = "bold"),
legend.position = "none", # Hide legend
strip.text = element_text(size = 16), # Facet label text
panel.grid.major = element_blank(), # Remove major grid lines
axis.text.x = element_blank(),
axis.text.y = element_text(angle = 90, vjust = 0.5, hjust = 1), # Rotate x-axis text
axis.title.y = element_text(size = 30)
) +
ylim(0,100) +
labs(
title = " ",
x = " ",
y = "Ratings"
) +
facet_grid(~variable, scales = "free_x", space = "free") +
scale_color_manual(values = c("music" = "red", "ambiguous" = "blue", "non-music" = "#3CB371")) +
scale_fill_manual(values = c("music" = "red", "ambiguous" = "blue", "non-music" = "#3CB371"))
ratings_cluster_plot_mean
Radar Chart of mean feature ratings for different clusters (Figure 3C)
audio_descriptives_wo_monra <- audio_descriptives %>%
dplyr::select(cluster,label, mean_repetition, mean_timbre, mean_tempo, mean_pulse, mean_harmony, mean_melody, mean_rhythm, mean_complexity, mean_intentionality, mean_instrumental)
subgroup_summary_pag <- audio_descriptives_wo_monra %>%
group_by(label) %>%
summarise(across(starts_with("mean"), list(mean = mean), .names = "{.col}_{.fn}")) %>%
rename(mean_pulse = mean_pulse_mean,
mean_rhythm = mean_rhythm_mean,
mean_timbre = mean_timbre_mean,
mean_repetition = mean_repetition_mean,
mean_tempo = mean_tempo_mean,
mean_melody = mean_melody_mean,
mean_harmony = mean_harmony_mean,
mean_intentionality = mean_intentionality_mean,
mean_instrumental = mean_instrumental_mean,
mean_complexity = mean_complexity_mean)
# Prepare the data by adding max and min rows
max_values <- rep(100, ncol(subgroup_summary_pag) - 1)
min_values <- rep(0, ncol(subgroup_summary_pag) - 1)
data_for_radar <- rbind(max_values, min_values, subgroup_summary_pag[, -1])
# Add row names for clarity (optional)
rownames(data_for_radar) <- c("Max", "Min", paste(subgroup_summary_pag$label))
data_for_radar <- data_for_radar %>%
rename(Repetition = mean_repetition,
Timbre = mean_timbre,
Instrumental = mean_instrumental,
Intentionality = mean_intentionality,
Complexity = mean_complexity,
Melody = mean_melody,
Harmony = mean_harmony,
Rhythm = mean_rhythm,
Pulse = mean_pulse,
Tempo = mean_tempo)
# Create the radar chart
radarchart(data_for_radar,
axistype = 1,
# Customize the appearance
pcol = c("blue", "red", "green"), # Colors for each cluster (nonmusic, ambiguous, music)
plwd = 4, # Line width
plty = 1, # Line type
cglcol = "grey", # Color of the grid lines
cglty = 1, # Type of the grid lines
axislabcol = "black", # Color of the axis labels
caxislabels = seq(0, 100, by = 20), # Axis labels from 0 to 100
vlcex = 1.5, # Variable label size
calcex = 0.9, # Axis label size
maxmin = TRUE, # Ensure max and min values are used
seg = 5 # Force 5 segments (0, 20, 40, 60, 80, 100)
)
Now, the idea is to try and identify different participant profiles to underIdentifying participant classes
suppressWarnings(suppressMessages(library(broom)))
suppressWarnings(suppressMessages(library(dplyr)))
suppressWarnings(suppressMessages(library(ggfortify)))
suppressWarnings(suppressMessages(library(gt)))
suppressWarnings(suppressMessages(library(jtools)))
suppressWarnings(suppressMessages(library(lcmm)))
suppressWarnings(suppressMessages(library(lme4)))
suppressWarnings(suppressMessages(library(lmerTest)))
suppressWarnings(suppressMessages(library(lmtest)))
suppressWarnings(suppressMessages(library(stringr)))
suppressWarnings(suppressMessages(library(viridis)))
suppressWarnings(suppressMessages(library(xtable)))We carry out a latent class linear mixed model to identify different participant profiles. We try having up to 6 different profiles (This code will be time consuming, it took us half a day to run.)
Notice that we use “class” and “profile” interchangeably…
m_random <- hlme(
fixed = monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
random = ~ 1,
subject = 'exp_subject_id',
ng = 1,
data = ratings
)
m_2cl <- hlme(
fixed = monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
mixture = ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
random = ~ 1,
subject = 'exp_subject_id',
ng = 2,
B = m_random, # Use m_random as initial values
data = ratings
)
m_3cl <- hlme(
fixed = monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
mixture = ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
random = ~ 1,
subject = 'exp_subject_id',
ng = 3,
B = m_random, # Use m_random as initial values
data = ratings
)
m_4cl <- hlme(
fixed = monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
mixture = ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
random = ~ 1,
subject = 'exp_subject_id',
ng = 4,
B = m_random, # Use m_random as initial values
data = ratings
)
m_5cl <- hlme(
fixed = monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
mixture = ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
random = ~ 1,
subject = 'exp_subject_id',
ng = 5,
B = m_random, # Use m_random as initial values
data = ratings
)
m_6cl <- hlme(
fixed = monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
mixture = ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse,
random = ~ 1,
subject = 'exp_subject_id',
ng = 6,
B = m_random, # Use m_random as initial values
data = ratings
)
d<-summarytable(m_random, m_2cl, m_3cl, m_4cl, m_5cl, m_6cl,
which = c("G", "loglik", "npm", "BIC", "%class", "entropy")) #??? classes is better
d<- as.data.frame(d)
#view(d)
summary(m_random)
summary(m_2cl)
summary(m_3cl)
summary(m_4cl)
summary(m_5cl)
summary(m_6cl)
logLik_2cl <- m_2cl$loglik
logLik_3cl <- m_3cl$loglik
logLik_4cl <- m_4cl$loglik
logLik_5cl <- m_5cl$loglik
logLik_6cl <- m_6cl$loglik
Selection of the number of classes: 7 criteria (see https://www.displayr.com/work-number-classes-latent-class-analysis/#:~:text=There%20are%20seven%20approaches%20to,your%20own%20Latent%20Class%20Analysis!)
We will look at the BIC and the “no small classes” criteria
No small classes: The basic idea of this approach is that you choose the highest number of classes, such that none of the classes are small (e.g., less than 5% of the sample). This rule has long been used in practice as a part of the idea of domain-usefulness but has recently been discovered also to have some theoretical justification (Nasserinejad, K., van Rosmalen, J., de Kort, W., Lesaffre, E. (2017) Comparison of criteria for choosing the number of classes in Bayesian finite mixture models. PloS one, 12).
Based on these criteria, and table “d”, we picked the solution with 5 classes.
postprob(m_2cl)
postprob(m_3cl)
postprob(m_4cl)
postprob(m_5cl)
postprob(m_6cl)
summary(m_5cl)
people_2_classes <- as.data.frame(m_2cl$pprob)
people_3_classes <- as.data.frame(m_3cl$pprob)
people_4_classes <- as.data.frame(m_4cl$pprob)
people_5_classes <- as.data.frame(m_5cl$pprob)
people_6_classes <- as.data.frame(m_6cl$pprob)
# View(people_5_classes)
people_2_classes$twoclasses <- people_2_classes$class
people_3_classes$threeclasses <- people_3_classes$class
people_4_classes$fourclasses <- people_4_classes$class
people_6_classes$sixclasses <- people_6_classes$class
Here, I add the 5-class-solution in the ratings DF.
str(people)
#Class is the solution we picked, it includes 5 classes, numbers from 1 to 5 shows which class a participants has been assigned.
ratings_class <- ratings %>%
left_join(people_5_classes %>%
dplyr::select(exp_subject_id, class),
by = "exp_subject_id")
The mean ratings of features for each of the 5 profiles.
cluster_means <- aggregate(
. ~ class,
data = ratings_class[, c("class", "intentionality", "instrumental", "complexity", "repetition", "melody", "harmony", "rhythm", "timbre", "tempo", "pulse")],
mean
)
cluster_sd <- aggregate(
. ~ class,
data = ratings_class[, c("class", "intentionality","instrumental", "complexity", "repetition", "melody", "harmony", "rhythm", "timbre", "tempo", "pulse")],
sd
)
print(cluster_means)
print(cluster_sd)
Now we carry LMMs for each profile. Caution, that we have less participants in each model, increasing the possibility for multicollinarity and other issues…
##1 CLASS
class1_models <- ratings_class %>%
filter(class == "1") %>%
lmer(monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse + (1 | exp_subject_id) + (1 | audio_name), data = .)
summary(class1_models)
r.squaredGLMM(class1_models)
car::vif(class1_models)
rand(class1_models)
performance::icc(class1_model, by_group = TRUE)
class2_models <- ratings_class %>%
filter(class == "2") %>%
lmer(monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse + (1 | exp_subject_id) + (1 | audio_name), data = .)
summary(class2_models)
r.squaredGLMM(class2_models)
car::vif(class2_models)
rand(class2_models)
performance::icc(class2_models, by_group = TRUE)
class3_models <- ratings_class %>%
filter(class == "3") %>%
lmer(monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse + (1 | exp_subject_id) + (1 | audio_name), data = .)
summary(class3_models)
r.squaredGLMM(class3_models)
car::vif(class3_model)
rand(class3_model)
performance::icc(class3_model, by_group = TRUE)
class4_models <- ratings_class %>%
filter(class == "4") %>%
lmer(monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse + (1 | exp_subject_id) + (1 | audio_name), data = .)
summary(class4_models)
r.squaredGLMM(class4_models)
car::vif(class4_model)
rand(class4_model)
performance::icc(class4_model, by_group = TRUE)
class5_models <- ratings_class %>%
filter(class == "5") %>%
lmer(monra ~ intentionality + instrumental + complexity + repetition + melody + harmony + rhythm + timbre + tempo + pulse + (1 | exp_subject_id) + (1 | audio_name), data = .)
summary(class5_models)
r.squaredGLMM(class5_models)
car::vif(class5_model)
rand(class5_model)
performance::icc(class5_model, by_group = TRUE)Reported figure in the paper
# Generate Viridis color palette with 5 distinct colors
viridis_colors <- viridis_pal(option = "D")(5)
plot_summs(class1_models, class2_models, class3_models, class4_models, class5_models,
scale = FALSE,
colors = viridis_colors,
model.names = c("Class 1", "Class 2", "Class 3", "Class 4", "Class 5")) +
theme_minimal(base_size = 14) +
theme(
axis.text = element_text(size = 25),
legend.text = element_text(size = 16),
legend.title = element_blank(), # Adjust legend title size
plot.title = element_text(size = 20, face = "bold"),
axis.title.x = element_text(size = 20),
axis.title.y = element_blank()
) +
labs(color = "Class") # Rename the legend title
Participant characteristics and classes… We check if participants in these different profiles differ regarding age, auditory imagery, musical sophistication, or openness to experience…
means_scales_age_class <- means_scales_age %>%
left_join(people_5_classes %>%
dplyr::select(exp_subject_id, class),
by = "exp_subject_id")
str(means_scales_age_genres)
# ANOVA for 'bais'
anova_bais <- aov(bais ~ factor(class), data = means_scales_age_class)
summary(anova_bais)
# ANOVA for 'gold'
anova_gold <- aov(gold ~ factor(class), data = means_scales_age_class)
summary(anova_gold)
# ANOVA for 'total_openness'
anova_openness <- aov(total_openness ~ factor(class), data = means_scales_age_class)
summary(anova_openness)
# ANOVA for 'age'
anova_age <- aov(age ~ factor(class), data = means_scales_age_class)
summary(anova_age)
They don’t.
Reported figure of age, BAIS-V, Gold-MSI and Openness scores of all 5 classes
means_scales_age_class$class <- as.factor(means_scales_age_class$class)
ggplot(means_scales_age_class, aes(x = class, y = age, color = class, fill = class)) +
geom_boxplot(alpha = 0.5, position = position_dodge(width = 0.75)) +
geom_jitter(alpha = 0.05, size = 1) +
scale_color_viridis_d() +
labs(x = "Class", y = "Age") +
scale_fill_viridis_d() +
theme_minimal(base_size = 16) +
theme(
axis.title = element_text(size = 30),
plot.title = element_text(size = 18, face = "bold"),
legend.position = "none",
strip.text = element_text(size = 16),
panel.grid.major = element_blank(),
axis.text.y = element_text(angle = 90, vjust = 0.5, hjust = 1)
) +
ggplot(means_scales_age_class, aes(x = class, y = bais, color = class, fill = class)) +
geom_boxplot(alpha = 0.5, position = position_dodge(width = 0.75)) +
geom_jitter(alpha = 0.05, size = 1) +
scale_color_viridis_d() +
labs(x = "Class", y = "BAIS-V") +
scale_fill_viridis_d() +
theme_minimal(base_size = 16) +
theme(
axis.title = element_text(size = 30),
plot.title = element_text(size = 18, face = "bold"),
legend.position = "none",
strip.text = element_text(size = 16),
panel.grid.major = element_blank(),
axis.text.y = element_text(angle = 90, vjust = 0.5, hjust = 1)
) +
ggplot(means_scales_age_class, aes(x = class, y = gold, color = class, fill = class)) +
geom_boxplot(alpha = 0.5, position = position_dodge(width = 0.75)) +
geom_jitter(alpha = 0.05, size = 1) +
scale_color_viridis_d() +
labs(x = "Class", y = "Gold-MSI") +
scale_fill_viridis_d() +
theme_minimal(base_size = 16) +
theme(
axis.title = element_text(size = 30),
plot.title = element_text(size = 18, face = "bold"),
legend.position = "none",
strip.text = element_text(size = 16),
panel.grid.major = element_blank(),
axis.text.y = element_text(angle = 90, vjust = 0.5, hjust = 1)
) +
ggplot(means_scales_age_class, aes(x = class, y = total_openness, color = class, fill = class)) +
geom_boxplot(alpha = 0.5, position = position_dodge(width = 0.75)) +
geom_jitter(alpha = 0.05, size = 1) +
scale_color_viridis_d() +
labs(x = "Class", y = "Openness") +
scale_fill_viridis_d() +
theme_minimal(base_size = 16) +
theme(
axis.title = element_text(size = 30),
plot.title = element_text(size = 18, face = "bold"),
legend.position = "none",
strip.text = element_text(size = 16),
panel.grid.major = element_blank(),
axis.text.y = element_text(angle = 90, vjust = 0.5, hjust = 1)
)
Let’s summarize the findings in a table!
# Create a list of your models
models <- list(class1 = class1_models,
class2 = class2_models, class3 = class3_models,
class4 = class4_models, class5 = class5_models)
fixed_effects_list <- purrr::map(models, ~ broom::tidy(.x, effects = "fixed"))
# Combine into a single data frame
fixed_effects <- bind_rows(fixed_effects_list, .id = "class")
fixed_effects <- fixed_effects %>%
mutate(significance = case_when(
p.value < 0.001 ~ "***",
p.value < 0.01 ~ "**",
p.value < 0.05 ~ "*",
TRUE ~ "ns"
))
# Calculate R^2 for each model and include class
rsquared_values <- purrr::imap(models, ~ {
r2 <- r.squaredGLMM(.x) # Calculate R2
tibble(
class = .y, # Add class name (model name or identifier)
R2m = r2[1], # Marginal R2
R2c = r2[2] # Conditional R2
)
}) %>% bind_rows() # Combine the list of tibbles into a single data frame
# Print the resulting data frame
print(rsquared_values)
print(rsquared_values)
# Reshape to make features (terms) rows and classes columns
significance_table <- fixed_effects %>%
dplyr::select(class, term, significance) %>%
pivot_wider(names_from = class, values_from = significance, values_fill = "ns")
rsquared_rows <- rsquared_values %>%
pivot_longer(cols = starts_with("R"), names_to = "term", values_to = "value") %>%
mutate(term = case_when(
term == "R2m" ~ "Marginal R²",
term == "R2c" ~ "Conditional R²"
)) %>%
pivot_wider(names_from = class, values_from = value, names_prefix = "class") %>%
mutate(across(-term, as.character)) # Ensure compatibility with `significance
rsquared_rows <- rsquared_rows %>%
mutate(across(starts_with("class"), ~ as.character(.)))
rsquared_rows_clean <- rsquared_rows %>%
rename_with(~ sub("^class", "", .), starts_with("class"))
# Then you can continue your pipeline with the cleaned data
combined_table <- bind_rows(significance_table, rsquared_rows_clean)
participant_counts <- tibble(
term = "# of Participants",
class1 = "17",
class2 = "14",
class3 = "21",
class4 = "35",
class5 = "11"
)
# Append the participant counts to your original table
final_table <- bind_rows(combined_table, participant_counts)
# Print the updated table
print(final_table)
final_table %>%
mutate(across(
starts_with("class"),
~ ifelse(term %in% c("Marginal R²", "Conditional R²", "# of Participants"), ., str_sub(., 1, 4))
)) %>%
mutate(across(
starts_with("class"),
~ ifelse(term %in% c("Marginal R²", "Conditional R²"), as.numeric(.), .)
)) %>%
gt() %>%
tab_header(
title = "Significance of Predictors and R² Values Across Classes",
subtitle = "Significance levels: *** (<0.001), ** (<0.01), * (<0.05), ns (not significant)"
) %>%
data_color(
columns = starts_with("class"),
colors = scales::col_factor(
palette = c("yellow", "orange", "red", "white"),
domain = c("***", "**", "*", "ns")
)
) %>%
fmt_number(
rows = term %in% c("Marginal R²", "Conditional R²"),
decimals = 3
) %>%
tab_style(
style = cell_fill(color = "white"),
locations = cells_body(rows = term %in% c("Marginal R²", "Conditional R²"))
) %>%
tab_style(
style = cell_text(color = "black"),
locations = cells_body(rows = term %in% c("Marginal R²", "Conditional R²", "# of Participants"))
) %>%
tab_style(
style = cell_fill(color = "white"),
locations = cells_body(rows = term %in% c("# of Participants"))
)
library(writexl)
# Save the underlying final_table to Excel. We edited the table format and have that version in the paper (also with rounded values)
write_xlsx(final_table, path = "final_table.xlsx")
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